FIG. 1 illustrates a block diagram of transmitter/receiver ends using an OFDMA scheme in an uplink direction. First, data stream sent to users are digitally modulated using modulation techniques such as Quadrature Phase Shift Keying and 16 Quadrature Amplitude Modulation. After the modulation, a constellation mapping is performed on the modulated data streams which are then passed through serial-to-parallel converter and converted into Nu number of parallel symbols. Here, Nu represents a number of subcarriers allocated to a mobile station (MS). From a total of Nc number of subcarriers, these symbols are mapped to Nu number of subcarriers while remaining subcarriers (Nc−Nu) are mapped to zero, or put differently, the remaining subcarriers are padded (e.g., zero padding). Here, Nc represents a total number of subcarriers before a cyclic prefix is added. That is, the remaining Nc−Nu number of subcarriers are zero padded and then is applied Nc-point Inverse Fast Fourier Transform (IFFT).
Furthermore, in order to reduce inter-symbol interference, a cyclic prefix is added to the symbols and passed through a parallel-to-serial converter, which is then transmitted to channels. The symbols transmitted via channels are mapped to subcarriers in amount of Np+Nc according to IFFT and the cyclic prefix.
As illustrated in FIG. 1, the procedures of an OFDMA receiver are same as that of the transmitter except in reverser order.
FIGS. 2a-2c illustrate methods of mapping Nu number of subcarriers out of Nc total number of subcarriers according to conventional art. FIG. 2a illustrates a random allocation of subcarriers, FIG. 2b illustrates allocating the subcarriers by collecting the subcarriers in specified frequency bands, and FIG. 2c illustrates allocating each subcarrier throughout the entire frequency bands in equal intervals.
Since the mapping methods illustrated in FIGS. 2a-2c make use of the entire frequency bands, frequency diversity can be achieved. However, because each subcarrier is allocated individually, timing synchronization of OFDM symbol of different users can be off, and signal quality can suffer due to nearby subcarriers of different users if Doppler frequency is large. Furthermore, in the conventional OFDMA scheme, a single user uses a plurality of subcarriers and as a result, Peak-to-Average Power Ratio (PAPR) characteristics can get worse.
The OFDMA signals in a time domain comprises a large number of subcarriers modulated independently. Consequently, if these subcarriers are added on the same phase, a problem of a large size signal being generated can occur. Even though average power remains fixed at a certain level, a maximum power can increase drastically which in turn increases the PAPR. If the PAPR is increased, Analog-to-Digital Conversion (ADC) and Digital-to-Analog Conversion (DAC) becomes more complex while efficiency of a Radio Frequency (RF) power amplifier is reduced. In order to resolve these types of PAPR heating problem, several resolution encoding methods have been proposed including using signal distortion or a special forward error correction symbol set (excluding OFDM symbol) having a large PAPR.
In order to improve the PAPR characteristics, as illustrated in FIG. 3, a DFT spread OFDMA scheme has been proposed. FIG. 3 is a block diagram illustrating transmitting/receiving ends using a DFT spread OFDMA scheme.
The difference between the DFT spread OFDMA scheme and the conventional OFDMA scheme is that in the DFT spread OFDMA, Nu number of data symbols are Nu-point DFTed. Thereafter, as illustrated in FIG. 2c, the converted data symbols are mapped in equal intervals to the entire Nc number of subcarriers. Compared to the conventional OFDMA, the DFT spread OFDMA has improved PAPR characteristics.
Because data in the DFT spread OFDMA system are allocated in the frequency domain, it is necessary to perform channel estimation on the channels of the frequency domain. In the conventional OFDMA system, data and pilot are simultaneously allocated in the frequency domain, in the DFT spread OFDMA, the data is spread via the DFT matrix. As such, if the pilot is simultaneously processed with the remaining data, the spreading operation is also performed on the pilot, making channel estimation difficult.